作者机构：
[黄德修; 刘德明; 何军; 李蔚; 杨春勇] Dept. of Optoelectron. Eng., Huazhong Univ. of Sci. and Technol., Wuhan 430074, China;[李源] Dept. of Comp. Sci., Central China Normal Univ., Wuhan 430070, China
通讯机构：
[Li, W.] D;Dept. of Optoelectron. Eng., Huazhong Univ. of Sci. and Technol., China
期刊：
Lecture Notes in Computer Science,2004年3332:197-204 ISSN：0302-9743
通讯作者：
Liu, HY
作者机构：
[Liu, HY] Cent China Normal Univ, Dept Comp Sci, Wuhan 430079, Hubei, Peoples R China.;Wuhan Univ, Dept Finance, Sch Business, Wuhan 430072, Hubei, Peoples R China.
通讯机构：
[Liu, HY] C;Cent China Normal Univ, Dept Comp Sci, Wuhan 430079, Hubei, Peoples R China.
摘要：
An advanced content-based news video browsing and retrieval system, NewsBR, is proposed in this work. The system is built on high-accuracy news story segmentation and topic caption text extraction. Its main features include category-based news story browsing, key-frame-based video abstract and keyword-based news story retrieval. In the paper, news story segmentation and topic caption text extraction, as well as content-based video browsing and retrieval, are addressed in detail. The system is helpful and effective for the overall understanding of the news video content.
作者机构：
[金聪; 彭嘉雄] State Key Laboratory of Education, Huazhong Univ. of Sci. and Technol., Min. Image Proc./Intelligent Contr., Wuhan 430074, China;[金聪] Department of Computer Science, Central China Normal University, Wuhan 430079, China
通讯机构：
[Jin, C.] S;State Key Laboratory of Education, Huazhong Univ. of Sci. and Technol., Min. Image Proc./Intelligent Contr., China
作者机构：
[Shengsheng Yu] College of Computer Science and Technology, Huazhong University of Science and Technology, Wuhan, China;Department of Computer Science, Central China Normal University, Wuhan, China;[Jingzhong Mao] Department of Mathematics, Central China Normal University, Wuhan, China;[Kaihua Xu] Educational Informatization Research Center, Central China Normal University, Wuhan, China;[Yuhua Liu] College of Computer Science and Technology, Huazhong University of Science and Technology, Wuhan, China<&wdkj&>Department of Computer Science, Central China Normal University, Wuhan, China
通讯机构：
[Yuhua Liu] C;College of Computer Science and Technology, Huazhong University of Science and Technology, Wuhan, China<&wdkj&>Department of Computer Science, Central China Normal University, Wuhan, China
关键词：
maximum flow;minimum cutset;capacity-expanded network;optimization algorithm
摘要：
The paper points out the relationship between the bottleneck and the minimum cutset of the network, and presents a capacity expansion algorithm of network optimization to solve the network bottleneck problem. The complexity of the algorithm is also analyzed. As required by the algorithm, some virtual sources are imported through the whole positive direction subsection in the network, in which a certain capacity value is given. Simultaneously, a corresponding capacity-expanded network is constructed to search all minimum cutsets. For a given maximum flow value of the network, the authors found an adjustment value of each minimum cutset arc’s group with gradually reverse calculation and marked out the feasible flow on the capacity-extended networks again with the adjustment value increasing. All this has been done repeatedly until the original topology structure is resumed. So the algorithm can increase the capacity of networks effectively and solve the bottleneck problem of networks.
作者机构：
[Tan, LS; Chen, L; Yin, M] Cent China Normal Univ, Dept Comp Sci, Wuhan 430079, Peoples R China.
通讯机构：
[Tan, LS] C;Cent China Normal Univ, Dept Comp Sci, Wuhan 430079, Peoples R China.
会议名称：
International Conference on Communication Technology
会议时间：
APR 09-11, 2003
会议地点：
BEIJING, PEOPLES R CHINA
会议主办单位：
Cent China Normal Univ, Dept Comp Sci, Wuhan 430079, Peoples R China.
关键词：
ABR service;high-speed computer networks;congestion control;PID controller;performance evaluation
摘要：
In this paper, we present a novel control-theoretic explicit rate (ER) allocation algorithm for congestion control of available bit rate (ABR) service in high-speed computer communication networks. The proposed control uses a proportional-integrative-plus-derivative (PID) controller rather than the usually adopted proportional-plus-derivative (PD) controller. Based on a general traffic model of computer network and on system stability criterion, it is shown that under PID controller the source rates are regulated in such a manner that the congestion-controlled network is asymptotic stable in terms of both the buffer occupancy of the destination node and the user transmission rates, and the bandwidth fairness is achieved. The basic control theory approach for the algorithm is firstly presented, specifically z-transformation and Schur-Cohn stability test are used to achieve the stability. Steady state analysis is subsequently given to show how the max/min fairness is achieved in a natural way without additional computation. We then use simulations to compare the performance of the PID congestion control scheme with the PD control scheme under a variety of networking conditions and loads. Our simulation results show that certain PID scheme produces advantages over PD scheme, namely PID control scheme leads to less buffer occupancy, faster response of the buffer occupancy as well as of the controlled ABR rates, smaller steady overshoots of ABR traffic and better utilization of network links.
期刊：
PACLIC 17: Language, Information and Computation, Proceedings,2003年:282-289
作者机构：
[He, TT; Xiaohua, LH; Liu, JZ] Cent China Normal Univ, Dept Comp Sci, Wuhan 430079, Peoples R China.
摘要：
Automatic Multi-word Units Extraction is an important issue in Natural Language Processing. This paper has proposed a new statistical method based on a large-scale balanced corpus to extract multi-word units. We have used two improved traditional parameters: mutual information and log-likelihood ratio, and have increased the precision for the top 10,000 words extracted through the method to 80.13%. The results of the research indicate that this method is more efficient and robust than previous multi-word units extraction methods.
作者机构：
[Liu, YH] Huazhong Univ Sci & Technol, Coll Comp Sci & Technol, Wuhan 430074, Peoples R China.;Cent China Normal Univ, Dept Comp Sci, Wuhan 430079, Peoples R China.;Cent China Normal Univ, Dept Math, Wuhan 430079, Peoples R China.;Cent China Normal Univ, Educ Informat Res Ctr, Wuhan 430079, Peoples R China.
关键词：
maximum flow;minimum cutset;capacity-expanded network;optimization algorithm.
摘要：
The paper points out the relationship between the bottleneck and the minimum cutset of the network, and presents a capacity expansion algorithm of network optimization to solve the network bottleneck problem. The complexity of the algorithm is also analyzed. As required by the algorithm, some virtual sources are imported through the whole positive direction subsection in the network, in which a certain capacity value is given. Simultaneously, a corresponding capacity-expanded network is constructed to search all minimum cutsets. For a given maximum flow value of the network, the authors found an adjustment value of each minimum cutset arc's group with gradually reverse calculation and marked out the feasible flow on the capacity-extended networks again with the adjustment value increasing. All this has been done repeatedly until the original topology structure is resumed. So the algorithm can increase the capacity of networks effectively and solve the bottleneck problem of networks.
摘要：
图像分割是任何图像分析过程中的首要任务，因为接下来所要做的工作(如特征提取、目标识别等）都取决于图像分割的质量。文章提出了一种改进的区域生长彩色图像分割方法iRGA(improved Region Growing Approach)--基于子块色彩均值和方差的图像分割法。实验证明，该方法可以有效地减少计算复杂度和抑制噪声的干扰。
期刊：
PACLIC 17: Language, Information and Computation, Proceedings,2003年:235-242
作者机构：
[He, TT; Luo, CR] Cent China Normal Univ, Dept Comp Sci, Wuhan 430079, Peoples R China.
摘要：
In this paper, vector-distance-weighted algorithm and representative-vector-distance algorithm are described and used to implement the process of automatic text classification. Two experiments have been done by means of the algorithms (experiment1 is based on vector-distance-weighted algorithm and experiment2 is based on representative-vector-distance algorithm). Characters are selected as features. The average precision of experiment1 and experiment2 is 80.36% and 69.27%, respectively. Comparing the two experiments, it can be concluded that the efficiency of text classification can be improved by means of vector-distance-weighted algorithm.
摘要：
To guarantee network Quality of Service (QoS), network controllers designed and applied in real networks are necessary. How to automatically find efficient parameters of a controller and get the setting time of control system in simulation is a common problem in the process of controller design. Some definitions, experience rules and algorithms based on statistical theory are proposed in this paper, which are able to get the setting time, judge the stability feature of network control system and compare the control effects of different controllers. We tested the algorithms with network congestion controllers and the experiments results show that they are efficient to network controller evaluation, which enlarge the testing range and facilitate controller design.
